JP2008204133A - Answer search apparatus and computer program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an answer search apparatus that answers a question by using related concepts. <P>SOLUTION: The answer search apparatus 44 includes a case frame dictionary group 40 including case frame records describing semantic relationships between declinable words and nominal concepts, a question reception part 100, a matching case frame search part 102 for searching the case frame dictionary group 40 for records whose nominative is a word in a question, a case frame selection part 108 for selecting the combination of a declinable word and an objective noun that has the highest occurrence frequency in the records found, a similar case frame search part 110 for searching a case frame dictionary user-selected from the case frame dictionary group 40 for records having the selected combination, a word selection part 116 for selecting the nominative noun included in the records found that has the highest occurrence frequency, and an output creation part 120 for generating an answer to the question with the use of the selected noun. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an apparatus for searching for an answer to a question by a user, and more particularly to an answer searching apparatus for searching an appropriate answer that is easy for a user to understand for a simple question.

  One of the functions provided by a computer is a search function for searching desired data from a large amount of data. A typical search system is one in which one or a plurality of keywords are input, and a document including all or any of the keywords is searched and displayed. However, in such a system, since a document is searched based on whether or not a keyword is included separately from the inquirer's intention, there is a problem that a document desired by the user cannot be selected accurately. is there.

  As one means for solving such a problem, there is a method of inputting a question sentence in a natural language, performing language analysis, and searching for a document based on the analysis result. However, even with such a method, a document is basically searched based on whether or not a certain word is included in the document, so that the same problem as that when using a keyword occurs.

  Furthermore, considering that there are a plurality of readings for a certain concept, there is a method of searching for a synonym using a thesaurus for an input keyword and searching for a document using the result. However, in this case, there is a problem that the number of documents to be searched is further increased and an accurate document cannot be presented.

One method for solving such a problem is disclosed in Patent Document 1 described later. In the system disclosed in Patent Document 1, a search process is executed based on an input question, and initial answer candidates are extracted based on the result. A query related to each answer candidate is retrieved by performing a query using the extracted answer candidates as keywords. It is determined whether or not the constituent words of the sentence included in the related document have the same vocabulary meaning as the constituent words of the input question, and the answer candidate determined to have the same vocabulary meaning is used as the input question. Select as an answer to.
JP 2006-293830 A

  However, in the system disclosed in Patent Document 1 described above, a sentence that seems to be appropriate is simply selected as an explanation for the input question, and from the viewpoint of how to present the input question to the user in an easy-to-understand manner. No solution has been made. Some questions are not about commentary on an inputted question, but seek a topic around the question. The system disclosed in Patent Document 1 cannot answer such a question.

  That is, for some questions, rather than explaining the question itself, it presents the surrounding concepts around the question in terms of semantics and concepts, and explains such concepts or things. May be more important. In the prior art, answer retrieval from such a viewpoint has not been performed.

  Therefore, an object of the present invention is not to simply provide an explanation for the question itself, but to provide an answer search apparatus capable of answering a question using peripheral concepts centered on the question.

  The answer search device according to the first aspect of the present invention includes a first case frame record in which a predicate, a noun concept that can be semantically connected to the predicate, and a case relationship between them are described. And a first storage means for storing the second case frame dictionary, a question input means for receiving an input of a question consisting of nouns, and a case frame record having a word inputted by the question input means as a main case A first search means for searching in the first case frame dictionary stored in the first storage means; a case frame record searched by the first search means; A first selection means for selecting a case frame record searched by the first search means with the highest appearance frequency from a combination of a noun that is a target case for the word, and a first selection; Selected by means A second search means for searching for a case frame record having a combination of a word and a noun in a second case frame dictionary stored in the first storage means; A second selection means for selecting a predetermined number of the most frequently occurring nouns in the main frame contained in the case frame record, and a question using the predetermined number of main nouns selected by the second selection means Answer output means for generating and outputting an answer to.

  In this answer search device, when the question input means receives an input of a question, the first search means searches the first case frame dictionary for a case frame record whose main case is a word included in the question. . The first selection means selects the most frequently searched combination from the combination of the prescription included in the searched case frame record and the target noun (if any). . The second search means searches the second case frame dictionary for a case frame record having this combination as a combination of a predicate and an object. A predetermined number of nouns having the highest appearance frequency are selected by the second selection means from the nouns that are the main case in the searched case frame records. The answer output means generates and outputs an answer to the question using the predetermined number of main nouns.

  Rather than searching directly for documents containing nouns contained in the question for the entered question and making it an answer, the commonly used utterances associated with the noun and its purpose (if any) A case frame record having the combination is searched for in the second case frame dictionary using the combination having the highest frequency in the case frame dictionary. A predetermined number of nouns with the highest frequency are finally selected for the main nouns in the case frame records searched in this way, and used for generating an answer. Since nouns that are highly related to the word in the question are searched and used in the answer using the noun and the noun that serves as the objective, create an answer from another aspect, not a direct answer to the question Can do. That is, various answers can be obtained to the user's question. As a result, it is possible to provide an answer search device that can answer a question using a peripheral concept centered on the question.

  Preferably, the first storage means stores a plurality of area-specific frame dictionaries respectively obtained from documents having different target reader areas, and the answer search device further includes a plurality of answer search devices based on a profile of a user who inputs a question. Means for selecting, as a second case frame dictionary, a region case frame dictionary obtained from a document in the target reader region that matches the profile among the region case frame dictionaries.

  As the second case frame dictionary, an area-specific case frame dictionary obtained from a document in the target reader area that matches the user profile is used. Therefore, the word presented to the user as an answer becomes familiar to the user, and an easier-to-understand answer can be generated.

  Preferably, the second selection means includes means for selecting one noun having the highest appearance frequency from the main nouns included in the case frame record searched by the second search means, and the answer output The means uses the noun included in the question and one noun selected by the second selection means, and generates and outputs an answer explaining the noun contained in the question by a metaphor using one noun. Including means.

  For a word included in a question, a combination of frequently used prescriptions and objective case nouns is obtained from the first case frame dictionary, and the combination is similarly used as a medium. In addition, another noun that is frequently used as a main case is selected from the second case frame dictionary. The selected noun is considered to be particularly semantically related to the word in question, and the word in question can be rephrased by another word that is closely related. At this time, by using a word paraphrase as a metaphor and using it as an answer, it is possible to generate an answer in a form that is easier for the user to understand.

  The second selection means includes means for selecting a plurality of nouns from the most frequently occurring nouns in the main case nouns included in the case frame record searched by the second search means, and outputs an answer The means may include means for generating and outputting an answer indicated as an alternative candidate for the noun included in the question using a noun included in the question and a plurality of nouns.

  A noun with a high frequency of use and purpose that is used at the same time as the word in the input question is obtained, and a plurality of nouns that are frequently used as a main case are obtained. All of these are considered to be deeply related to the words in the question, and by presenting a plurality of these to the user, the user is given an option for the noun included in the question. It becomes possible to select what the user desires from these, and an answer useful for the user can be provided.

  Preferably, the answer search device includes: a second storage unit for storing a thesaurus including synonyms for the nouns and nouns; a synonym for the nouns included in the combination selected by the first selection unit; A synonym search means for searching a synonym of a noun that is a target case for a word by means of a thesaurus, a synonym for a predicate searched by the synonym search means, and a synonym of a noun and a case having the combination. And third search means for searching for a frame record in the second case frame dictionary stored in the first storage means, wherein the second selection means includes the second search means and the third search means. Frequency order selection means for selecting a predetermined number of the most frequently occurring nouns in the main case included in the case frame record searched by the search means. .

  The second case frame dictionary not only frequently uses the nouns in the question as the main case but also the synonyms of the synonyms and the nouns that are frequently used as the main case. You can search from The range of noun candidates to be searched becomes wider, and the possibility of obtaining a more appropriate answer increases.

  More preferably, the frequency order selection means sets the appearance frequency of the main nouns searched for by the second search means among the main nouns included in the case frame records searched by the second search means and the third search means. On the other hand, in order to select the predetermined number of main nouns having the highest appearance frequency by calculating the sum of the appearance frequencies of the nouns of each major case by relatively reducing the appearance frequency of those searched by the third retrieval means. Including means.

  A combination including a synonym has a certain relationship to the question, but is considered to have a lighter relationship to the question than the original combination. Therefore, it is reasonable to give a smaller weight to the frequency of nouns searched for combinations including synonyms than the original combination when searching for an answer to a question. By using such weights, it is possible to expand the search range and increase the number of candidate nouns, while reducing the possibility that important nouns will be buried in other less important nouns and leaked from the search results. Can do. As a result, it is possible to provide an answer search device that can give an appropriate answer to a question.

  The computer program according to the second aspect of the present invention is a computer program that, when executed by a computer, causes the computer to function as each of the above-described answer search means.

  By causing the computer program to be executed by the computer, the computer functions as each means of any one of the answer search apparatuses described above. As a result, the computer can obtain the same effect as the above-described answer search device.

[First Embodiment]
FIG. 1 is a block diagram showing a schematic configuration of a phrase explanation system 30 based on a metaphor, which is an answer search apparatus according to the first embodiment of the present invention. Referring to FIG. 1, this system 30 includes a case frame dictionary including a plurality of case frame dictionaries which are databases that store a number of structures called case frames indicating the relationship between prescriptions and nouns related to the predicates. A group 40, a thesaurus 42 in which synonyms are hierarchically grouped for Japanese words, an input / output device 46 in which a user inputs a question and an answer from the system 30 is displayed, and an input / output device 46 For a given question, the case frame dictionary group 40 and the thesaurus 42 are used to generate a phrase explanation by metaphor for the question, and a phrase explanation apparatus 44 by metaphor for presenting to the questioner via the input / output device 46. Including.

  “Case frame” is used in the field of natural language processing to express the semantic structure of verbs, adjectives and adjective verbs (“Iwanami Information Science Dictionary” (Nagao et al. Edit, Iwanami Shoten, published in 1990), page 109). The “case frame” has several “case slots”, and the “case slot” represents contents such as an actor case (main case), a target case (target case), and a place case.

  In the present embodiment, as shown in FIG. 1, the question screen 50 includes an input field 60 for entering an input word and a question button 62 operated by the user to give the word as a question to the phrase explanation device 44 using a metaphor. Including. By inputting a word to be a question in the input field 60 and pressing a question button 62, the word input in the input field 60 is given to the phrase explanation device 44 by a metaphor as a question. The phrase explanation device 44 generates a phrase explanation based on a metaphor using the case frame dictionary group 40 and the thesaurus 42 for this word, and gives it to the input / output device 46.

  The phrase explanation screen 52 based on metaphor displayed on the input / output device 46 includes a field 70 in which explanation is displayed, and a question button 72 operated by the user to ask the next question. When the question button 72 is pressed, the screen changes to the question screen 50.

  The case frame dictionary group 40 is a database composed of tables including a plurality of case frame records. “Case frame record” refers to one unit in a database table constituting a case frame dictionary. The configuration of the case frame record in this embodiment will be described later. Each case frame record generally indicates a relationship structure between a predicate and a noun associated with the predicate. Referring to FIG. 2, for example, the verb “represent” takes “president” as the main character and “country” as the target character as one example. In the case of Japanese, the particles “ha” and “wo” representing the main case and the purpose case are added, respectively. Similarly, the verb “to govern” can have “president” as the main character and “country” as the target character. In the case frame record of the case frame dictionary group 40, one verb, the main noun for the verb obtained from the example, and the target noun, if any, are stored as one set. In general, many case frame records are stored in the case frame dictionary for one verb.

  Case frame dictionaries can be created not only for verbs but also for adjectives or adjective verbs. For example, for the adjective "Shibui", "Ginkakuji" and "Engenji" can be considered as semantic principals. In fact, if the examples of “Ginkakuji are sober” and “Enkai are sober” existed in the knowledge sources when creating case frame dictionaries, these case frame headings are also included in case frame dictionaries. Will be stored. The same is true for adjective verbs.

  There may also be a case frame structure without an objective case. For example, the words “smiling” and “astringent” can be taken as the main characters of both “Ginkakuji” and “enshi” as shown in FIG. In this case, as shown in FIG. 2, case structures corresponding to these two examples are stored in the case frame dictionary.

  Referring to the lower part of FIG. 2, case frame record 80 of the case frame dictionary in the present embodiment is specific to part of speech (“representative”), part of speech label (for example, “verb”), and part of speech. Data structure. The data structure peculiar to the part of speech is a data structure that is defined separately according to each part of speech because the case frame structure that can be taken is different for each of the verb, the adjective, and the adjective verb. For example, in the case of a verb, if it takes a subject and an object, the first related word (eg “President”), its case label (eg “Main case”), its part of speech (eg “Noun”), A particle attached to the word (eg “ha”), a second related word (eg “country”), a case label (eg “objective case”), a part of speech (eg “noun”), Including particles (for example, “O”).

  The same data structure can be considered for adjectives and adjective verbs. However, adjectives and adjective verbs have a simpler structure than verbs.

  In the present embodiment, a plurality of case frame dictionaries are prepared as the case frame dictionary group 40. Each of these case frame dictionaries is created from knowledge sources in an appropriate field according to, for example, the technical field, the age group of the questioner, and the occupation. For example, if the target knowledge layer is an elementary school student, it is only necessary to collect examples from sentences appearing in textbooks used in elementary schools and create a case frame dictionary. Similarly, for a technical field, a case frame dictionary may be created by collecting examples from a collection of papers in the technical field.

  There are various case frame dictionary structures, but in this embodiment, as shown in FIG. 2, the relationship in sentence structure between a prescriptive word and a noun related to the prescriptive word can be understood. It is only necessary to have a data structure that is in a format and can restore the original example.

  FIG. 3 shows in what format a set of a prescription constituting a headline and an example corresponding to the prescription is stored for the contents of a case frame dictionary 90 included in the case frame dictionary group 40. Is illustrated. Referring to FIG. 3, for example, the record relating to the first prescription V1 is obtained from an example of the form “X1 V1 N1”, and is shown in the case frame record 80 of FIG. Has a similar structure. The same applies to the term V2. The word V1 and the word V2 may coincide with each other.

  FIG. 4 is a block diagram showing a detailed configuration of the phrase explanation device 44 based on the metaphor shown in FIG. Referring to FIG. 4, phrase explanation device 44 matches question accepting processing unit 100 for accepting a questioner's question (word) from input / output device 46, and the question word accepted by question accepting processing unit 100. A matching case frame search unit 102 for searching for an entry having a word as a main case from the case frame dictionary group 40, and a matching case frame storage for storing headings of case frames searched by the matching case frame search unit 102 Part 104. In the present embodiment, the matching case frame search unit 102 selects an appropriate case frame dictionary that matches the user profile from the case frame dictionary group 40 based on the user profile input in advance, and performs a search. Shall be used.

  Here, the “case frame dictionary matching the user profile” refers to a case frame dictionary preliminarily assigned as appropriate to the category when the user is classified into a plurality of categories based on the items in the user profile. I mean. For example, users can be classified by age or school year. In this case, a case frame dictionary created from a document for a child, student or student of that age or grade can be prepared, and an appropriate case frame dictionary can be selected according to the value of the age or grade item in the user profile. When classifying users according to their occupations, case frame dictionaries created from specialized field documents can be prepared and appropriate ones can be selected according to the occupations of the users. Furthermore, users can be classified according to a plurality of items in the user profile. In that case, a case frame dictionary is created from the documents classified in advance according to the plurality of items, and an appropriate one is selected according to the combination of the values of the plurality of items in the user profile. it can.

  “Metaphor” refers to borrowing and expressing each thing that has the same characteristics as each other when explaining something. In particular, when explaining things that are not familiar, it is said that it is easier for the listener to understand if expressed in a metaphor using familiar things.

  The word explanation device 44 further includes a frequency calculation processing unit 106 for calculating the appearance frequency of records that match the case frame records stored in the matching case frame storage unit 104 except for the main case part, and a matching case frame A case frame selection processing unit 108 for selecting and outputting the record with the highest frequency calculated by the frequency calculation processing unit 106 among the records stored in the storage unit 104 is included.

  The phrase explanation device 44 further includes a case frame record having a prescription and an objective case that match the prescription and the objective case included in the case frame record selected by the case frame selection processing unit 108, and the prescription and the objective case. Similar case frame search processing unit 110 for retrieving and outputting records having synonyms of nouns as predicates and objective cases from the case frame dictionary group 40, and records retrieved by the similar case frame search processing unit 110 A main word storage unit 112 for storing the main word included in the main word, a frequency calculation processing unit 114 for calculating the appearance frequency of the main word stored by the main word storage unit 112, and a main word storage unit 112. Selection processing unit for selecting and outputting the word with the highest frequency calculated by the frequency calculation processing unit 114 among the words stored in And a 16. Note that the similar case frame search processing unit 110 obtains synonyms of the word of the predicate and the objective case from the thesaurus 42.

  The phrase commentary device 44 further receives from the questioner the fixed expression storage unit 118 for storing the fixed expression used in the explanation by the metaphor, the word selected by the word selection processing unit 116, and the question reception processing unit 100. An output creation process for generating a description using a metaphor for the question word based on the question word and the case frame record selected by the case frame selection processing unit 108 and outputting the comment to the input / output unit 46 (see FIG. 1). Part 120.

  FIG. 5 schematically shows what records are searched by the phrase explanation device 40 and how metaphors are generated, taking the question “President” (what is it?) As an example. Referring to FIG. 5, when the word “President” is input, the matched case frame search unit 102 selects the input question word from the case frame dictionary selected according to the user profile in the case frame dictionary group 40. A case frame record having a matching noun as a main case is searched (process 132). An example of the case frame record searched by the search processing 132 is shown in block 134 of FIG. Subsequently, among these case frame records, a process 136 is performed for calculating the appearance frequency of each of the cases where the predicate coincides with the object, and selecting the one with the highest appearance frequency. Here, it is assumed that the combination of a noun with a noun of “the (president) governs the country” and an objective noun is selected.

  Subsequently, the case frame record having the same combination of the noun and the noun of the objective case as the combination of the noun of the selected case frame record and the noun of the objective case is obtained by the similar case frame search processing unit 110 shown in FIG. A process 139 for retrieving from the dictionary is executed. An example of a case frame record retrieved by this process is shown in block 140. Similarly, the similar case frame search processing unit 110 uses the thesaurus 42 to obtain synonyms for each of the nouns of the case frame record 138 selected by the case frame selection processing unit 108 and the noun of the target case. A process 144 for searching a case frame record having a synonym as a predicate and a noun of the target case from the case frame dictionary is executed. The resulting case frame record group is shown in block 146.

  The case frame record group shown in block 140 and the case frame record group shown in block 146 are stored separately in the main word storage unit 112 shown in FIG.

  Subsequently, the frequency calculation processing unit 114 calculates the appearance frequency of the main word for the case frame records shown in the block 140 and the block 146 (processing 148). At this time, the frequency calculation is performed with the weight = 1 for the word appearing in the block 140 and the weight = 0.7 for the main word appearing in the block 146.

  The word selection processing unit 116 selects the main word 150 having the highest frequency as a result of the frequency calculation performed with weighting in this way, and outputs the selected main word 150 to the output creation processing unit 120. In this example, it is assumed that the word 150 “king” is selected as a result of this processing.

  The output creation processing unit 120 inserts a word (for example, “President”) input as a question sentence into the blank portion 154 of the portion representing the subject in the fixed expression 152 stored in the fixed expression storage unit 118. Further, the word 150 selected by the word selection processing unit 116 in the blank part 158 (for example, “king”), the noun of the case frame record used in the case frame record selected by the case frame selection processing unit 108 in the blank part 156 Each of the expressions (for example, “rule the country”) is inserted, and the output screen 50 shown in FIG. 1 is created and given to the input / output device 46.

  As a result, for example, a questioner who has entered the question “What is the president?” Can obtain a metaphorical explanation that the president is like a king in the sense of governing the country. In this example, since it is assumed that the questioner is an elementary school student, the case frame dictionary used for the search in the similar case frame search processing unit 110 is a dictionary for elementary school students selected from the case frame dictionary group 40 according to the user profile. Use. By doing so, an answer to the question can be generated with a metaphor using words that are easy to understand for the user in a form corresponding to the user profile.

  As will be described later, the phrase explanation device 44 according to the present embodiment can be realized by computer hardware and a computer program executed on the computer hardware. As will be described later, FIG. 6 shows a computer program control structure for realizing the word explanation device 44 according to this embodiment in a flowchart format.

  Referring to FIG. 6, this program is started when the user inputs a word in input field 60 and presses question button 62 on question screen 50 shown in FIG. 1. At this time, the word input in the input field 60 is passed as an argument to this program.

  This program includes a step 180 for determining which case frame dictionary in the case frame dictionary group 40 is to be used from a user profile recorded in advance, and a verb having the input question word as the main case. In step 182 for searching in the dictionary, in step 184 for calculating the appearance frequency of case frame records having the same prescription and noun of the same objective case among the case frame records searched in step 182; Selecting and storing the case frame record having the highest frequency among the calculated frequencies.

  The program further retrieves from the case frame dictionary selected to match the user profile a case frame record that contains the same word and object as the case frame record found in step 186. Step 187, searching for the synonym of the prescription of the case frame record selected in step 186 from the thesaurus 42, and synonym of the noun that is the object of the case frame record selected in step 186 from the thesaurus 42 A case frame record having a combination of synonyms of the search and step 188 and the synonyms of the word and the object searched in steps 188 and 190 is searched from the case frame dictionary selected in step 180 so as to match the user profile. And the accumulated steps including the main word And a flop 192.

  The program further includes a step 194 for calculating the appearance frequency for each word that is the main case appearing in the case frame record accumulated in step 192. In step 194, the case frame record searched in step 187 is assigned a weight of 1, the case frame record searched in step 192 is assigned a weight of 0.7, and the frequency is calculated to calculate the sum thereof.

  The program further includes a step 196 for selecting a noun which is the main word calculated in step 194, a word selected in step 196, and a question word passed as an argument when the program is started. Step 198 creates a metaphorical commentary by inserting these words into a pre-stored fixed expression using the noun of the case frame record selected in step 186 and the noun of the target case. And a step 200 for outputting the created explanatory text to the input / output device 46. When step 200 ends, the program ends.

  The phrase explanation device 44 according to the first embodiment can be realized by executing a program having a control structure as shown in FIG. 6 by computer hardware.

[Second Embodiment]
FIG. 7 is a schematic block diagram of a tourist facility recommendation system 230 which is an answer search device according to the second embodiment of the present invention. Referring to FIG. 7, this tourist facility recommendation system 230 includes a case frame dictionary group 40, a thesaurus 42, and phrases used in the first embodiment similar to those used in the first embodiment. A tourist facility recommendation device 244 having a configuration similar to that of the explanation device 44 and an input / output device 46 connected to the tourist facility recommendation device 244 are included.

  Referring to FIG. 7, this tourist facility recommendation system 230 is for a user to input a certain word and recommend a facility related to the word or a similar facility. As shown in FIG. 7, the input screen 250 in the system 230 includes an input field 260 for inputting a facility name, and a recommendation button 262 that is clicked when the user recommends the facility. On the input screen 250, a sentence “Where do you want to go?” Is displayed in advance, and an input field 260 is provided following this sentence. Since the word “like” is displayed after the input field 260, it is easy for the user to know that the name of some place or facility is input in the input field 260.

  As shown in FIG. 7, the output screen 252 of the system 230 includes a field 270 that lists recommended destinations, and a button 272 that the user operates to receive a recommendation of a similar facility by listing another facility. .

  FIG. 8 is a schematic block diagram of the tourist facility recommendation device 244 shown in FIG. 8, blocks having the same functions as those in FIG. 4 are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated.

  The tourist facility recommendation device 244 shown in FIG. 8 differs from the phrase commentary device 44 shown in FIG. 4 in that the tourist facility is illustrated using the input screen 250 shown in FIG. 7 instead of the question acceptance processing unit 100 shown in FIG. 4 is included in the case frame records stored in the main word storage unit 112 instead of the word selection processing unit 116 shown in FIG. Including the word selection processing unit 302 for selecting a predetermined plurality from the top ones with the highest frequency, and a plurality selected by the word selection processing unit 302 instead of the output creation processing unit 120 shown in FIG. And an output creation processing unit 306 for creating and outputting the output screen 252 shown in FIG. 7, and the standard sentence for facility recommendation stored in the regular expression storage unit 118.

  The question reception processing unit 300 has the same function as the question reception processing unit 100 shown in FIG. 4 except that the input screen 250 shown in FIG. 7 is used instead of the question screen 50 shown in FIG.

  The word selection processing unit 302 shown in FIG. 8 is the same as the word selection processing unit 116 shown in FIG. 4 except that a plurality of words are selected from the highest frequency instead of the word selection processing unit 116 shown in FIG. It has the function of.

  Unlike the output creation processing unit 120 shown in FIG. 4, the output creation processing unit 300 shown in FIG. 8 creates the output screen 252 shown in FIG. 7 using a plurality of words given from the word selection processing unit 302. .

  FIG. 9 shows a flowchart of a computer program for realizing the tourist facility recommendation device 244 according to the second embodiment. The program shown in FIG. 9 is started when the user inputs some value in the input field 260 and clicks the recommendation button 262 on the input screen 250 shown in FIG. This program is given the input text in the input field 260 as a parameter.

  In FIG. 9, the same steps as those shown in FIG. 6 are denoted by the same reference numerals. The processing performed in these steps is the same as the processing executed in the corresponding steps in FIG. Therefore, detailed description thereof will not be repeated here.

  9 differs from FIG. 6 in that steps 350, 352 and 354 are included instead of steps 196, 198 and 200 shown in FIG.

  In step 350 shown in FIG. 9, words appearing as main characters are sorted in descending order of the frequency calculated in step 194. Subsequently, in step 352, a recommendation sentence (recommendation screen) as shown in the output screen 252 of FIG. 7 is created using a plurality of words determined in advance from the head of the sorted main word list. In step 354, the recommendation screen generated in this way is output to the input / output device 46.

  In the apparatus according to the second embodiment, for example, when a user inputs a place such as “museum” and requests a recommendation of a tourist facility, the museum is designated as the main character in the case frame dictionary group 40. All the case frame records of the predicates that have as are searched. Of the case frame records searched in this way, the most frequently used combination is selected from the combination with the word as the target case if it is a predicate. Further, the case frame dictionary selected according to the user profile in the case frame dictionary group 40 has the selected combination and the combination of the synonym of the prescriptive word and the synonym of the target case noun. Case frame records are searched. Of these case frame records, the appearance frequency of the main noun is calculated. In this case, different weights are used for the combination of the word of the prescription and the objective case selected by the case frame selection processing unit 108 and the case frame record obtained by using the synonym searched by the thesaurus 42. After calculating the appearance frequency of the main word in this way, the word selection processing unit 302 sorts the list of words in descending order according to the frequency, selects a predetermined plurality from the highest frequency, and generates an output generation processing unit To 306. The output creation processing unit 306 creates an output screen 252 shown in FIG. 7 using these words.

  In the output screen 252, it is convenient for the user to embed a link to a web page or the like describing the facility in more detail in the name of the recommended tourist facility. For this purpose, for a tourist facility that appears in the case frame dictionary, a database that holds the URL (Universal Resource Locator) is created in advance, and when the output screen 252 is created in the output creation processing unit 306, the URL What is necessary is just to give the link to to the text of the name of each tourist facility.

  In this way, using the mechanism almost the same as that of the first embodiment, the tourist facility is not based on the inputted word, but the device for explaining the phrase using the metaphor as in the first embodiment. A device to recommend can be realized. In this process, a predicate (and a combination of words as a target case) having the input tourist facility name as a main word is searched, and a main word having the prescription (and a combination of target words) is further searched. The search is performed from the case frame dictionary selected according to the user profile. Among the searched word groups, a plurality of words having the highest appearance frequency are recommended to the user.

  By doing so, the names of other tourist facilities that have a deep relationship with the predicate are searched, centering on the premise that is conceptually deeply related to the facility name input by the user. As a result, it is possible to select a tourist facility having properties similar to the tourist facility preferred by the user and present it to the user.

  In the above-described embodiment, in the case of the search by the similar case frame search processing unit 110, the case frame dictionary group 40 used for the search by the matching case frame search unit 102 is matched with the user profile. Used. However, the present invention is not limited to such an embodiment. For example, a case frame dictionary completely different from the case frame dictionary group 40 may be used in the search by the similar case frame search processing unit 110.

  Further, in the above embodiment, the database describing the semantic relationship between the prescription and the noun is read as “case frame dictionary”, and its form is as shown in FIGS. However, the present invention is not limited to such an embodiment. Any type of database may be used as long as it is a database that understands the relationship between a predicate, a noun that is a principal for the predicate, and a noun that is a target case (if any).

  Moreover, in the said embodiment, the word is only accepted as a question. This is because it is possible to easily search for answers by doing so. However, if a word bearing the central meaning of a sentence is extracted by analyzing the sentence structure, it is possible to receive the question sentence itself as an input instead of the word.

<Realization by computer>
FIG. 10 executes the computer program having the control structure shown in FIGS. 6 and 9 for the phrase commentary system 30 according to the first embodiment and the tourist facility recommendation system 230 according to the second embodiment, respectively. FIG. 11 shows an internal configuration of the computer system 530 in a block diagram form.

  Referring to FIG. 10, a computer system 530 includes a computer 540 having a memory drive 552 and a DVD (Digital Versatile Disc) drive 550, a keyboard 546, a mouse 548, a monitor 542, a microphone 570, and a pair of speakers. 572.

  Referring to FIG. 11, in addition to the memory drive 552 and the DVD drive 550, the computer 540 boots up a CPU (Central Processing Unit) 556, a bus 566 connected to the CPU 556, the memory drive 552, and the DVD drive 550, and A read-only memory (ROM) 558 for storing programs and the like, and a non-volatile memory connected to the bus 566 and a random access memory (RAM) 560 for storing program instructions, system programs, work data, and the like, and a bus 566 And a hard disk 554 which is an external storage device.

  Computer 540 further includes a network adapter 576 that provides a connection to a local area network (LAN) 574.

  A computer program for causing the computer system 530 to operate as the word commentary system 30 or the tourist facility recommendation system 230 is stored in the DVD 562 or the non-volatile memory 564 inserted into the DVD drive 550 or the memory drive 552, and further the hard disk 554. Forwarded to Alternatively, the program may be transmitted to the computer 540 through the network 574 and stored in the hard disk 554. The program is loaded into the RAM 560 when executed. The program may be loaded directly into the RAM 560 from the DVD 562, from the nonvolatile memory 564, or via the network 574.

  The case frame dictionary group 40, the thesaurus 42, and the user profile shown in FIG. 1 and the like are stored on the hard disk 554 and loaded into the RAM 560 when the program is executed. The CPU 556 reads an instruction from an address on the RAM 560 indicated by a program counter register (not shown), decodes the instruction, reads data from an address specified by the decoding result of the RAM 560 or the hard disk 554, processes the instruction according to the instruction, and decodes the instruction. The processing result is stored at the address specified by the result. The CPU 556 repeats these processes to generate and present an answer to the question.

  The input / output device 46 shown in FIGS. 1 and 7 is realized by the monitor 542, the keyboard 546, the mouse 548, and the like shown in FIGS. As will be readily understood by those skilled in the art, the generated answer may be output as synthesized speech. In this case, the speaker 572 can be used as an output device.

  This program includes a plurality of instructions for causing the computer 540 to operate as the system 30 or 230 according to the above embodiment. Some of the basic functions required to perform this operation are provided by operating system (OS) or third party programs running on the computer 540 or various toolkit modules installed on the computer 540. Therefore, this program does not necessarily include all functions necessary for realizing the system of the above-described embodiment. This program includes only instructions for executing the operation as the system 30 or 230 described above by calling an appropriate function or “tool” in a controlled manner so as to obtain a desired result. It only has to be. Here, the “tool” refers to a relatively small program that causes a computer to perform routine processing suitable for a certain purpose. A “tool kit” refers to a set of programs in which many of these tools are collected and made available from the outside.

  The operation of computer system 530 is well known and will not be repeated here.

  The embodiment disclosed herein is merely an example, and the present invention is not limited to the above-described embodiment. The scope of the present invention is indicated by each claim in the claims after taking into account the description of the detailed description of the invention, and all modifications within the meaning and scope equivalent to the wording described therein are intended. Including.

It is a schematic block diagram of the phrase commentary system 30 by the metaphor which is an answer search apparatus which concerns on the 1st Embodiment of this invention. It is the schematic for demonstrating the concept of a case frame. It is the schematic for demonstrating the structure of a case frame dictionary. 4 is a block diagram showing a schematic configuration of a word commentary device 44. FIG. It is a figure for demonstrating operation | movement of the word commentary apparatus 44. FIG. It is a flowchart which shows the control structure of the computer program for implement | achieving the phrase description apparatus 44 by computer hardware. It is a schematic block diagram of the sightseeing institution recommendation system 230 which is an answer search apparatus concerning the 2nd Embodiment of this invention. FIG. 8 is a block diagram showing a schematic configuration of a tourist facility recommendation device 244 shown in FIG. 7. It is a flowchart which shows the control structure of the computer program for implement | achieving the tourist facility recommendation apparatus 244 by computer hardware. It is a figure which shows the general view of the computer system 530 which implement | achieves the system which concerns on the 1st and 2nd embodiment of this invention. FIG. 3 is a block diagram showing a hardware configuration of a computer system 530.

Explanation of symbols

30 Phrase Explanation System by Metaphor 40 Case Frame Dictionary Group 42 Thesaurus 44 Phrase Explanation Device by Metaphor 46 Input / Output Device 100, 300 Question Accepting Processing Unit 102 Matching Case Frame Searching Unit 104 Matching Case Frame Accumulating Unit 106, 114 Frequency Calculation Processing Unit 108 Case frame selection processing unit 110 Similar case frame search processing unit 112 Primary word storage unit 116, 302 Word selection processing unit 118 Fixed expression storage unit 120, 306 Output creation processing unit

Claims (7)

A first case for storing first and second case frame dictionaries including a number of case frame records describing a predicate, a noun concept that can be semantically connected to the predicate, and a case relationship between them. Storage means,
Question input means for receiving input of questions consisting of nouns;
First search means for searching in the first case frame dictionary stored in the first storage means for a case frame record having the word input by the question input means as a main case;
Of the combinations of the case frame records searched by the first search means and the nouns that are the target case for the words, if any, the case frame records searched by the first search means A first selection means for selecting the one having the highest appearance frequency,
A second case for searching in the second case frame dictionary stored in the first storage means for a case frame record having a combination of a predicate and a noun selected by the first selection means; Search means;
A second selection means for selecting a predetermined number having the highest appearance frequency among the main case nouns included in the case frame record searched by the second search means;
An answer search device comprising: an answer output means for generating and outputting an answer to the question using the predetermined number of main nouns selected by the second selection means. The first storage means stores a plurality of area-specific frame dictionaries respectively obtained from documents having different target reader areas,
The answer search device further includes, based on a user profile stored in advance, an area-specific frame dictionary obtained from a document in a target reader area that matches the profile among a plurality of area-specific frame dictionaries. The answer search apparatus according to claim 1, further comprising means for selecting as a case frame dictionary. The second selection means includes means for selecting one noun having the highest appearance frequency among the main nouns included in the case frame record searched by the second search means,
The answer output means uses the noun included in the question and the one noun selected by the second selection means, and the metaphor using the one noun as the noun included in the question. The answer search apparatus according to claim 1, further comprising means for generating and outputting an answer explained by. The second selection means includes means for selecting a plurality of nouns from the most frequently occurring nouns in the main case nouns included in the case frame record searched by the second search means,
The answer output means includes means for generating and outputting an answer indicated as an alternative candidate for the noun included in the question using the noun included in the question and the plurality of nouns. The answer search device according to claim 1 or 2. A second storage means for storing a thesaurus including synonyms for the nouns and nouns;
Synonyms for the terms included in the combination selected by the first selection unit, and synonyms for a noun that is the target case for the term, using the thesaurus,
For each combination of a synonym for the predicate searched by the synonym search means and a synonym of the noun, a case frame record having the combination is stored in the second case stored in the first storage means. A third search means for searching in the frame dictionary;
The second selection means is a frequency for selecting a predetermined number having the highest appearance frequency from the main nouns included in the case frame records searched by the second search means and the third search means. The answer search apparatus according to claim 1, comprising an order selection unit. The frequency order selection means determines the appearance frequency of the main nouns searched for by the second search means among the main nouns included in the case frame records searched by the second search means and the third search means. On the other hand, the appearance frequency of what is searched for by the third search means is converted to a relatively small value to calculate the sum of the appearance frequencies of the nouns of each main character, and a predetermined number of main nouns having the highest appearance frequency are selected. The answer search apparatus according to claim 5, further comprising means for: A computer program that, when executed by a computer, causes the computer to function as each means according to any one of claims 1 to 6.

Images